ISSN: 2277-9655 et al., IC™ Value: 3.00 CODEN: IJESS7 … /Archive-2017/September... · · 2017-09-25frequency domain watermarking systems are comparatively predominant than the

ISSN: 2277-9655

[Shankar * et al., 6(9): September, 2017] Impact Factor: 4.116

IC™ Value: 3.00 CODEN: IJESS7

http: // www.ijesrt.com© International Journal of Engineering Sciences & Research Technology

[611]

IJESRT INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES & RESEARCH

TECHNOLOGY AN OPTIMIZED DWT BASED APPROACH IN VIDEO WATERMARKING WITH

MULTIPLE WATERMARKS USING OPPOSITIONAL KRILL HERD

ALGORITHM T. Shankar*1 and G.Yamuna2

*1Assistant Professor, Department of ECE, Annamalai University, India 2Professor, Department of ECE, Annamalai University, India

DOI: 10.5281/zenodo.996033

ABSTRACT In video watermarking applications, there is a need to extract the watermark without using the original data

because of the huge storage of the cover data. In this paper, we have intended to present an efficient multiple

video watermarking using optimized three levels DWT. Basically, the system consists of three modules such as

(i) selecting optimal wavelet coefficients using Oppositional Krill Herd Algorithm (OKHA) (ii) Watermark

embedding process and (iii) Watermark extraction process. A hybrid combination of the Krill Herd with

Oppositional-Based Learning (OBL) has been used to select the optimal wavelet co-efficient OBL is used to

improvise the performance of Krill Herd algorithm, while optimizing the co-efficient of standard DWT model. It

is then followed by encryption of watermark based on Arnold transform. After, that we encrypt the watermark

based on the Arnold transform and different types of media (gray image and colour image) are used for the

watermark. Finally, the original video is obtained with the help of extraction process. The robustness of this

technique is tested by various attacks such as: noising, compression, and image-processing attacks

KEYWORDS: Video Watermarking, DWT, Oppositional Krill Herd Algorithm, Opposition-based learning,

multiple watermarking.

I. INTRODUCTION Digital watermarking is a data hiding method that is utilized for copyright fortification of digital media.

Dissimilar steganography, where the message is associated with the cover information [1, 2] copyrights data

information signify the owner rights known as ‘‘watermark’’ are entrenched into the digital media in a method

of image, audio, or video without distressing the perceptual quality. Imperceptibility and sturdiness in

contradiction of attacks are the basic problems in digital watermarking systems [3]. The watermark is

entrenched and abstracted as per prerequisite to signify the ownership and/or the individuality of multimedia [4].

In video watermarking, the watermark can be entrenched in the spatial domain [5] or in the transmute domain

such as Discrete Cosine Transform (DCT) [6], Discrete Fourier Transform (DFT) [7], and DWT [8]. The

frequency domain watermarking systems are comparatively predominant than the spatial domain watermarking

systems, in lossy compression, noise addition, rescaling, pixel removal, rotation and cropping.

A digital watermark is an indiscernible signal further to digital information, known as cover work that can be

perceived later for buyer/seller documentation, ownership proof, and so forth [9]. It shows the role of a digital

signature, as long as the image with an intellect of ownership or legitimacy. A watermark is proficient of

displaying frequent momentous features. These encompasses that the watermark is hard to observe, tolerates

common distortions, resists malicious attacks, conveys numerous bits of data, is capable of being

contemporaneous with other watermarks, and difficult to insert or detect [10]. In order for a watermark to be

valuable it must be robust to a diversity of conceivable attacks through pirates. These comprise robustness

against compression namely JPEG, scaling and aspect ratio changes, cropping, rotation, row and column

removal, addition of noise, cryptographic, filtering and statistical attacks, and also enclosure of other

watermarks [11].

In this association, a flood of numerous methods and modifications are efficiently engaged on the image to

obscure the associated data limited in the image. The transform domain embedding is well-defined as the

http://www.ijesrt.com/

ISSN: 2277-9655




[612]

domain of assimilating approaches for that a host of algorithms have been envisioned [12]. Additionally, the

important image watermarking methods include the Discrete Cosine Transform (DCT) [13], Discrete Wavelet

Trans-form (DWT), Z-transform [14], bit-plane compression techniques [15, 16], and the LSB based method

[17]. Furthermore, the wavelet-based watermarking system to entrench multiple watermarks in medical images

is enlightened. The watermarking methods for protected data hiding in wavelet trampled fingerprint images are

elucidated in [18, 19]. Visual Cryptography is additional field that is utilized for content protection. Visual

cryptography has been implemented to numerous applications, including but not limited to data hiding, overall

access edifices [20], visual confirmation and empathy [21].

In this paper, video approaches for multiple watermarks using oppositional krill Herd algorithm is developed.

Here (i) optimal three-level discrete wavelet transform (DWT) based watermark embedding/extracting, (ii)

Arnold transform based watermark encryption/decryption method and (iii) different types of watermarks

(gray/color) are used. The rest of this paper is organized as follows. Section II describes the details of the

literature survey and section III describing the proposed video watermarking scheme and the experimental

results being presented in Section IV and Section V summarizing the work.

II. RELATED WORKS A host of inquiries are accessible in literature associated with the topic of video watermarking, some of that are

reviewed at this time. Mohamed M. Ibrahimet al. [22] has enlightened the Video Multiple Watermarking

Technique Based on Image Interlacing using DWT. This method was on the basis of image interlacing. In this

method, three-level discrete wavelet transform (DWT) was utilized as a watermark embedding/extracting

domain, Arnold transform was utilized as a watermark encryption/decryption technique, and dissimilar kinds of

media (gray image, color image, and video) were utilized as watermarks. The sturdiness of this method was

tested by implementing dissimilar kinds of attacks namely: geometric, format-compression, noising, and image-

processing attacks.

Chitrasen and TanujaKashyap [23] persuasively scheduled the digital video watermarking engaging DWT for

the drive of Data Security. In their innovative method, the random segmentation and modernization of engrained

confidential information was performed without any preceding notion concerning the innovative host video. In

the sequence of the inserting process, the confidential information was engrained in discrete video frames

engaging the DWT’s frequency domains. A DWT video watermark technique was luminously brought in for the

drive of maintenance information. In addition to this, they efficiently projected a Haar wavelet based digital

watermark method to defend the information. In the Haar wavelet, Dyadic equation was engaged to graft the

copy right information in video bit streams.

In [24], Marwen Hasnaoui and Mihai Mitrea luminously transported to light the conceptual outline enabling the

binary Quantization Index Modulation (QIM) implanting approaches that had to be protracted in the direction of

multiple-symbol QIM. The uncomplicated detection method was augmented in relation to the minimization of

the average error probability under the semblance of white, additive Gaussian conduct for the assaults. In this

way, for the quantified parameters like the lucidity and robustness, the information payload was enriched by a

factor of log2m. M-QIM was legitimate based on inquiries against the edifice of the MEDIEVALS French

national project and of the SPY ITEA2 European project, related to the MPEG-4 AVC robust and semi-fragile

watermarking implementations.

Agilandeeswari and Ganesan [25] have elucidated an effectual rescindable adaptive video watermarking system

for multiple watermarks on the basis of Bi-directional Associative Memory (BAM) Neural Networks and Fuzzy

Inference System specifically, Multi-BAM-FUZ scheme. The chief aim of this paper was to design a robust

video watermarking scheme that simplifies secure video broadcast over a communication channel by preserving

a trade-off amongst imperceptibility, robustness and then the watermark capacity or payload. The BAM neural

network provisions, creation of weight matrix (formed out of multiple images) and this matrix was entrenched

into the DWT uncorrelated mid frequency coefficients of entirely the constituents (Y, Cb, Cr) of all frames of

the video with fluctuating entrenching strength ‘α’. This adaptive embedding strength was engendered with the

help of the Fuzzy Inference System that receipts HVS features namely luminance, texture and edge of each

frame as an input in the DWT transform.


ISSN: 2277-9655




[613]

Natarajan Mohananthini and Govindarajan Yamuna [26] presented a multiple medical image watermarking

scheme based on discrete wavelet transform (DWT) and singular value decomposition (SVD) to increase the

security of medical images and to preserve the privacy of patients.

The robust watermarking outline with a blind abstraction procedure for HEVC encoded video was provided by

Tanima Dutta and Hari [27]. A readable watermark was implanted invisibly in 4×4 intra prophesied blocks of

the HEVC encoded video. Their watermarking outline implements security by reconnoitring the spatio-temporal

features of the compressed video and a haphazard key for selection of inserting regions. They also analyze the

strengths of dissimilar compressed domain features of HEVC encoded video for applying the entrenching

algorithm. Experimental solutions was validate that their work confine the upsurge in video bit rate and

dilapidation of perceptual quality.

N.Mohananthini and G.Yamuna [28] proposed a multiple watermarking scheme based on DWT using genetic

algorithms to improve the performance of imperceptibility and robustness.

The QIM blind video watermarking system on the basis of Wavelet transform and principal constituent analysis

was elucidated by Nisreen et al. [29].The refuge of the system was reputable with the help of one secret key in

the recovery of the watermark. Discrete Wavelet Transform (DWT) was implemented on each video frame

disintegrating it into a number of sub-bands. Maximum entropy blocks were nominated and distorted by

Principal Component Analysis (PCA). Quantization Index Modulation (QIM) was utilized to quantize the

maximum coefficient of the PCA blocks of each sub-band. Then, the watermark was implanted into the

nominated appropriate quantize values. Their system was tested by a number of video systems.

III. PROPOSED VIDEO WATERMARKING METHODOLOGY The basic idea of proposed methodology is multiple video watermarking using oppositional krill herd algorithm

(OKHA). The proposed system consists of three modules such as (i) optimal wavelet coefficient selection, (ii)

Embedding process and (iii) Extraction process. Here, at first the input video is divided in to sub videos, among

the sub videos we select two identical copies of the original video. Depending on the sub videos the embedding

and extraction process are carried out. After that, we apply the three-level discrete wavelet transform (DWT) to

the selected video sub frames. In this approach, to increase the efficiency of the system the watermark images

also encrypted. The encrypted watermark image is embedded into the original sub video frame. The same

process is repeated for the extraction process also. The overall process of the proposed multiple video marking

system shows in figure 1.


ISSN: 2277-9655




[614]

Figure 1: Overall Diagram of Proposed Multiple Watermarking Scheme

IV. SELECTING OPTIMAL WAVELET COEFFICIENTS USING OKHA The objective of this section is to select the optimal coefficient using oppositional Krill Herd Algorithm. The

KHA algorithm is a meta-heuristic swarm intelligence optimization algorithm used to increase the PSNR and

NC values. Oppositional-Based Learning (OBL) is integrated with KHA for improving the convergence speed

and simulation results of conventional KH algorithm. A detailed proposal of optimal wavelet coefficient

selection using OKHA is explained below.

Step 1: Solution Encoding

The solution encoding is the important process in the optimization algorithm. In this work, we optimize the

wavelet coefficient of DWT. At first, we randomly assigned optimal values. These optimal values are used to

split the frame into sub bands. The frequency bands are selected depending on the range of the coefficient

values. The coefficient values vary with respect to lower bound ( iL ) and upper bound iU values. Initially set

the each coefficient with a real random value. To increase the image quality, the coefficient values are optimally

selected using OKHA.

]...,,[ 1 d

iii yyY Ni ,...,2,1 (1)

where, d

i

d

i

d

i ULy , , Dd ,...,2,1 , is the position of thi krill in the

thd dimension, D is the complete

dimension encoding with TNgNh independent variable in the search space and d

i

d

i UL , are the lower

and upper limits of thi krill in the

thd dimension.

Step 2: Opposite Solution of KHA

Rendering to opposition based learning (OBL) familiarized by Tizhoosh in 2005 [30], the current krill and its

opposite krill are measured instantaneously to attain a better calculation for present krill result. It is given that an

opposite krill solution has a better chance to be closer to the global optimal solution than random krill solution.

The opposite krill’s positions iOY are totally demarcated by constituents of iY .

]...,,[ 1 d

iii oyoyOY (2)

where, d

i

d

i

d

i

d

i yULoy with d

i

d

i

d

i ULoy , is the position of thi opposite agent iOY in the

thd

dimension of opposite population.

Step 3: Fitness Calculation


ISSN: 2277-9655




[615]

An objective function needs to be designed to evaluate the fitness of a solution and to quantify the performance

of each individual. In this research, the objective function is defined as follows:

PMaxfunctionFitnessH

(3)

MSEPSNRP

2

10

255log10 (4)

M

i

N

j

jiEjiENM

MSE0

2

0

,,1 (5)

where jiE , and jiE , are the corresponding pixel values in the original and watermarked frames and the

size of each frame is NM . It represents the coefficient range based on the equation (3), the ultimate goal is

maximization of the Objective function P ,which highly depends on filter coefficients. In the above

optimization problem, H represent the coefficient range.

Step 3: Calculation of Krill Movement

After the fitness calculation, the krill’s are moving to another place on the basis of the density performance.

Conferring to theoretical arguments, the krill individuals try to preserve a high density and move because of

their mutual effects [31]. The direction of motion inducedi , is estimated from the local swarm density (local

effect), a target swarm density (target effect) [33]. For a krill individual, this movement can be well-defined as:

old

ini

Maxnew

i MMM (6)

ett

i

local

ii

arg (7)

Where, maxM is the maximum induced speed, n is the inertia weight of the motion induced in the range

[0,1], old

iM is the last motion induced, local

i is the local effect provided by the neighbours and etT

i

arg is the

target direction effect provided by the best krill individual. In this study, the effect of the neighbors in a krill

movement individual is determined as follows:

NN

j

ijij

local

i XK1

ˆˆ (8)

ij

ij

ijXX

XXX̂ (9)

bestworst

ji

ijKK

KKK

ˆ (10)

where, bestK and

WorstK are the best and the worst fitness values of the krill individuals so far; iK signifies the

fitness or the objective performance value of the thi krill individual; jK is the fitness of

thj

NNj ,,.,2,1 neighbour; X signifies the related positions; and NN is the number of the neighbours. For

avoiding the singularities, a small positive number is added to the denominator. The sensing distance for each

krill individual can be determined with the help of the dissimilar heuristic approaches. At this time, it is dogged

for utilizing the subsequent formula for each iteration.

N

j

jiis XXN

D1

,5

1 (13)

where, isD , is the sensing distance for the thi krill individual and N is the number of the krill individuals. The

factor 5 in the denominator is empirically obtained. By means of equation (13), if the distance of two krill

individuals is less than the defined sensing distance, they are neighbours. The known target vector of each krill

individual is the lowermost fitness of an individual krill. The consequence of the individual krill with the best

fitness on the thi individual krill is considered using Equation (3). This level is communicated as

bestibesti

bestett

i XKC ,,

arg ˆˆ (14)


ISSN: 2277-9655




[616]

where, bestC is the effective coefficient of the krill individual with the best fitness to the

thi krill individual.

This coefficient is defined since ett

i

arg leads the solution to the global optima and it should be more operative

than other krill individuals such as neighbours. Herein, the value of bestC is defined as

max

12

IrandCbest (15)

Where rand is a random values between 0 and 1 and it is for enhancing exploration, I is the actual iteration

number and maxI is the maximum number of iterations.

Step 4 Calculation of Foraging Motion

The foraging motion is expressed in the name of two chief effective parameters. The primary one is the food

location and the second one is the preceding experience about the food location. This motion can be articulated

for the thi krill individual as follows:

old

ifIfi FWVF (16)

best

i

food

ii (17)

fV is the foraging speed, fW is the inertia weight of the foraging motion in the range [0,1], is the last foraging

motion, food

i is the best fitness of the thi krill so far. The centre of food for each iteration is formulated as

N

i i

N

i

i

ifood

K

XK

X

1

1

1

1 (18)

Therefore, the food attraction for the thi krill individual can be determined

foodifoodi

foodfood

i XKC ,,ˆˆ (19)

where foodC is the food coefficient. Because the effect of food in the krill herding decreases during the time, the

food coefficient is determined as

max

12I

IC food

(20)

The effect of the best fitness of the thi krill individual is also handled using the following equation:

ibestiibesti

Best

i XK ,,ˆˆ (21)

where, bestiK is the best previously visited position of the thi krill individual.

Step 5: Calculation of Physical Diffusion

The physical diffusion of the krill individuals is measured to be a haphazard procedure. This motion can be

expressed in terms of the maximum diffusion speed and a haphazard directional vector and can be articulated as

maxDDi (22)

where, maxD is the maximum diffusion speed, and the random directional vector and its arrays are random

values between -1 and 1.

Step 6: Updation based on KH Algorithm

Using different effective parameters of the motion during the time, the position vector of a krill individual

during the interval tt is given by the equation:

dt

dXttXttX i

ii (23)

Step 7: Termination Criteria

The algorithm discontinues its execution only if maximum number of iterations are achieved and the solution

which holding the best fitness value is selected and specified as best coefficient of wavelet. Once the best fitness

is attained by means of KHA algorithm, selected coefficient is allocated for DCT.


ISSN: 2277-9655




[617]

Embedding Process

Embedding is an essential method of converting the image into another format for video watermarking. This

encrypted watermark image is embedded into the original sub frame so as to withstand the malicious attacks on

the original information. The step by step process of proposed embedding process is explained below,

Step 1: Consider the input videoinV , which consist of n number of sub videos n

in VVVV ...,,, 21 . As first

step, the original video is interlaced into n number of sub videos Nn ,...,2,1 . The interlaced

process is explained in next section.

Step 2: The two most resembling sub videos ASub and BSub are picked from videoinV

. The resemblance is

deduced on the basis of normalized correlation (NC) measure which is a two step process. The average

NC value of all color sub bands and for all sub video frames is computed. Two sub videos that

constitute the biggest average NC are considered to be the most alike sub videos and these two videos

are agreed to the additional processing. The designated sub videos.

l

x

m

y

yxyx SSml

NC1 1

,.

1 (24)

where, yxS , and yxS , are corresponding pixel values of two sub frames and ml is the sub frame size.

Step 3: A three level discrete wavelet transform is applied to the selected sub video frames. Here, the optimal

wavelet coefficients are selected using OKHA algorithm, a detailed explanation of this algorithm is

lineated in the following section.

Step 4: After the selection of similar sub videos ASub and BSub , multiple watermarking scheme is performed.

In this paper, two types of watermark images such as gray image and colour image are used in

embedding process. Secret information is embedded into four parts of the original information between

sender and receiver.

Part 1: Multiple watermarks is different types

Part 2: Watermarked frame numbers corresponding to each watermark.

Part 3: Number of iterations applied to encrypt watermark image.

Part 4: Scaling factor for all sub video frames

Step 4: To improve the efficiency of the system before the embedding process, Arnold transform is applied to

encrypt the watermark imageinW .This number is the third part of the secret key between the sender

and the receiver.

Step 5: The watermark image is embedded into sub video A frames based on the below equation (25).

WII (25)

where, I → HL3 sub band of original sub video frame; I →Watermarked image

W → Watermark signal; → scaling factor

This scaling factor is a constant value for all subvideo frames. This scaling factor is the fourth and last

part of the secret key between the sender and the receiver.

Step 6: Finally, the watermarked video is generated by de-interlacing the watermarked subvideo and other sub

videos. This watermarked video and the secret key (of four parts) are sent to the receiver and no copy of the

original video is sent which is the ultimate goal of this paper.

Watermark Extraction Process

During the extraction process, an operation in reverse to the embedding process is performed, to extract the

watermark image inW from a subvideo and is compare with the original subvideo. In order to discover the

original sub video, in the beginning; the extraction algorithm performs the same set of operations as the

embedding algorithm. The watermarked video and four parts of secrete key are given as the input for the

extraction process. The step by step extraction process is explained below;

Step 1: Consider the watermarked video I , which is interlaced by the same level and type as the original video.

Step 2: Watermarked subvideo A and the subvideo B is selected by using the first part of the secrete key.

Step 3: The watermark extraction process is based on the selected sub videos.

The frame from both selected subvideos using the third part of the secret key is selected at first.

These frames are then transformed into frequency domain using three-level DWT.


ISSN: 2277-9655




[618]

Using the watermark extraction equation (26), the watermarks are extracted.

IIW (26)

where I is the HL3 subband of the watermarked subvideo A frame, I is the corresponding HL3 subband of

the subvideo B frame, W is the extracted encrypted watermark, and is the scaling factor from the fourth

part of the secrete key.

Number of Iterations is deduced from the second part of the secret key and inverse Arnold

Transform is applied that many iterations, to extract the encrypted watermark image.

Original video V is obtained based on the above process.

V. RESULTS AND DISCUSSION

The result of our proposed methodology is dissected in this section. The proposed watermarking scheme

discussed in this paper has effectively embedded the watermark image into video frame and extracted it back

from the embedded frame of the watermarked image. In our experiment, the standard video air horse is used.

Evaluation Metrics

The evaluation of the proposed technique is carried out using the following metrics as deducted by using below

equations

Peak Signal to Noise Ratio (PSNR)

The PSNR is used to measure the quality of the watermarked image. The PSNR is the ratio between the input

image jiI in , and the watermarked image jiI W , . The PSNR is identified using the mean square error

(MSE). The MSE gives the cumulative squared error between the corrupting noise and the maximum power of

the signal. Higher the PSNR and lower the MSE, better the quality of the image.

MSEPSNR

2

10

255log10

M

x

N

y

Win jiIjiINM

MSE1 1

2,,

1

where, jiI in , Input video frame jiI W , Watermarked frame

Arnold Transform

Arnold transform is one of the encryption algorithm [31] used to enhance the security and robustness of the

watermarking system. The Arnold transform uses the following formula for changing the location of each pixel

in the image/video frame of size NN .

y

x

y

x

y

x

21

11

where, yx, is the current location and yx , is the new location. Figure 6 shows the encryption of “Lena”

image using Arnold Transform. For a symmetric arnold transform, the number of iterations must be less than the

arnold period to used it as an encryption key. This technique is the image interlacing for image watermarking,

and the same technique can be used in videos but in two steps: first, the video frames are divided into subframes

using image interlacing and then these subframes are collected together to generate the subvideos.


ISSN: 2277-9655




[619]

(a) Original Image (b) Encrypted Image (c) Restored Image

Figure 2 Arnold Transform as Image / Video Frame Encryption Method

Interlacing Results

Image interlacing is an operation in which any image can be divided into sub-images was proposed by [32]. De-

interlacing is the reverse operation of combining sub-images together to generate the original image. In this

proposed method, a two level of interlacing is used. One-level interlacing has two types: interlacing by rows

only to get even rows (ER) and odd rows (OR) sub images and interlacing by columns only to get even columns

(EC) and odd columns (OC) sub-images. Two-level interlacing is a method of interlacing by rows first and then

by columns (or in the opposite order) to get four sub-images: even rows even columns (EE), even rows odd

columns (EO), odd rows even columns (OE), and odd rows odd columns (OO). The NC values are tabulated in

table 1.

Table 1: Video Interlacing Operation

Interlacing

Level Sub-videos Red Green Blue

One ER-OR 0.9532 0.9412 0.9234

EC-OC 0.9662 0.9542 0.9315

Two

EE-OE 0.9627 0.9231 0.9320

EE-EO 0.9615 0.9732 0.9645

EE-OO 0.9483 0.9265 0.9352

OE-EO 0.9650 0.9417 0.9712

OE-OO 0.9622 0.9755 0.9543

EO-OO 0.9654 0.9437 0.9367

Table1 shows the NC values between the two resulted subvideos in each interlacing level. From these results, it

is identified that all NC values are close to one, which means that the used interlacing algorithm gives subvideos

that are similar to each other. A closer comparison helps to deduce the most similar pair to be EC and OC for

one level interlacing and OE and OO for two-level interlacing and among color bands, red is found to be the one

which has the best result.

Simulation Results The main intention of proposed methodology is multiple video watermarking based on Oppositional Krill Herd

Algorithm (OKHA). The proposed method is tested with “Airhorse.avi” as a host video with, “Evil Inside.jpg”

as a grayscale image watermark of size 32 × 32 and “Lena.jpg” as a color image watermark of size 32 × 32 are

shown in figure 3. Three-level discrete wavelet transform is applied to the frame got by dissecting original

videos. Before apply the DWT, we calculate the optimal wavelet coefficient using Oppositional Krill Herd

Algorithm. Then, we perform the multiple watermark schemes. Finally, we obtain the watermarked video.


ISSN: 2277-9655




[620]

Original Video Frame (Gray) Grayscale Image

Watermark Watermarked Video Frame (Gray)

Extracted

Watermark

Original Video Frame (Color) Grayscale Image

Watermark Watermarked Video Frame (Color)

Extracted

Watermark

Figure 3 Original Video Frames and its Watermarked Video Frames

Table 2. Performance Comparison (Gray Image Watermark)

Gray Scale Image Watermark

M.Ibrahim et al. [22] Proposed Method

Interlacing Level No 1-Level 2 –Level No 1-Level 2 –Level

Sub-Videos EC-OC OE-OO EC-OC OE-OO

Scaling Factor 0.5 1.5 1.5 0.5 1.5 1.5

PSNR 36.7755 35.2069 33.9840 39.707 37.1348 35.3802

NC

No attacks 0.9824 0.9578 0.9725 0.9826 0.9580 0.9736

Gaussian Noise (10%) 0.6231 0.8794 0.8962 0.7650 0.8872 0.8971

Salt & Pepper Noise (10%) 0.8100 0.9382 0.9556 0.9432 0.9847 0.9881

Contrast Factor (10%) - - - 0.7781 0.8493 0.8412

JPEG Quality (80%) - - - 0.7232 0.8362 0.8372

JPEG Quality (60%) - - - 0.7351 0.8541 0.8554

Cropping 0.7250 0.7252 0.7303 0.7481 0.8423 0.8413


ISSN: 2277-9655




[621]

Table 3. Performance Comparison (Color Image Watermark)

Color Image Watermark

M.Ibrahim et al. [22] Proposed Method

Interlacing Level No 1-Level 2 –Level No 1-Level 2 –Level

Sub-Videos EC-OC OE-OO EC-OC OE-OO

Scaling Factor 0.5 1.5 1.5 0.5 1.5 1.5

PSNR 36.7755 35.2069 33.9840 39.7078 37.1348 35.3802

NC

No attacks 0.9824 0.9578 0.9725 0.9826 0.9580 0.9736

Gaussian Noise (10%) 0.6231 0.8794 0.8962 0.7650 0.8872 0.8971

Salt & Pepper Noise (10%) 0.8100 0.9382 0.9556 0.9432 0.9847 0.9881

Contrast Factor (10%) - - - 0.7781 0.8493 0.8412

JPEG Quality (80%) - - - 0.7232 0.8362 0.8372

JPEG Quality (60%) - - - 0.7351 0.8541 0.8554

Cropping 0.7250 0.7252 0.7303 0.7481 0.8423 0.8413

Table 2 shows the performance of the proposed method for multiple watermarks with attacks using air horse

video. In this proposed approach, we use the multiple watermarks such as gray image and color image

watermark. From table 2, it is inferred that the maximum PSNR of 39.70 is obtained for non-interlacing,

followed by 37.13 for using one-level interlacing and 35.38 for two-level interlacing. First, the results of the

with and without interlacing are close to each other which indicate that the goal of this paper is achieved.

Second, the results of both first and second level interlacing are close to each other which indicate that there is

no need for more levels of interlacing because there is no enhancement in results in two levels interlacing over

one level interlacing.. Moreover, we obtain the maximum normalized correlation value. During the

communications channel between the sender and the receiver, the watermarked video is attacked; these attacks

are categorized into five categories. These categories are as follows: geometric attacks, noising attacks, de-

noising (filtering) attacks, format compression attacks, and image-processing attacks. In order to ensure that our

proposed solution is not affected against the robustness of the video multiple watermarking systems against

attacks.

VI. CONCLUSION In this paper, a multiple watermarking approach based on three-level discrete wavelet transform and interlacing

technique is proposed. This method mainly reduces the non-blind watermarking system problems. In this

approach three-level discrete wavelet transform (DWT) was used as watermark embedding/extracting process

and Arnold transform as watermark encryption/decryption process. Proposed scheme has been tested against

different standard videos, and is inferred to get more efficient results. The results of the proposed watermarking

technique do not seem to be robust against attacks and results in higher manipulation of the watermark text after

extraction. In future we will try to enhance the performance of the proposed hybrid algorithm using different

strategy for embedding and extraction of watermark in order to impart better robustness to the scheme.

V. REFERENCES [1] P.C. Ritchey, V.J. Rego, “A Context sensitive tiling system for information hiding”, Journal of Inform

Hiding Multimedia Signal Process, vol. 3,no. 3, 2012.

[2] 2W.C. Kuo, “Image hiding by square fully exploiting modification directions”, Journal of Information

Hiding Multimedia Signal Process, vol. 4, no. 3, 2013

[3] M.K. Thakur, V. Saxena, J.P. Gupta, ‘‘A Performance analysis of objective video quality metrics for

digital video watermarking”, in: 3rd IEEE International Conference on Computer Science

andInformation Technology – ICCSIT 10, pp. 12–17, China, 9– 11,2010.

[4] P. Singh and R. S. Chadha, “A survey of digital watermarking techniques applications and attacks”,

International Journal of Engineering and Innovative Technology, vol.2, no.9, pp.165–175, 2013

[5] G. Langelaar, I. Setyawan, R. Lagendijk, “Watermarking digital image and video data”, IEEE Signal

Processing Magazine, vol.17, no. 9,pp.20–43, 2000.

[6] Han-Min Tsai, Long-Wen Chang, “Highly imperceptible video watermarking with the Watson’s

DCT-based visual model”, in: IEEE International Conference on Multimedia and Expo, ICME ‘04,

vol.3, pp. 1927–1930, 2004.


ISSN: 2277-9655




[622]

[7] M. Barni, F. Bartolini, R. Caldelli, A. De Rosa, A. Piva, “A Robust watermarking approach for raw

video, in: Proceedings 10th International Packet Video Workshop PV2000, Cagliari, Italy, 1–2, 2000.

[8] H. Liu, N. Chen, J. Huang, X. Huang, Y.Q. Shi, “A Robust DWT-based video watermarking

algorithm”, in: IEEE International Symposium on Circuits and Systems ISCAS 2002, vol. 3, pp. 631–

634, 26–29 May, 2002.

[9] Cox I. J., Miller, M. L. and Bloom J. A., “Digital Watermarking”, Morgan Kaufmann Publishers, USA,

2002.

[10] Miller, M.; Cox, I.J.; Linnartz, J.P.M.G.; Kalker, T., “A review of watermarking principles and

practices”, In Digital Signal Processing in Multimedia Systems, Edit. K. K. Parhi and T.

Nishitani,Marcell Dekker Inc., pp. 461-485, 1999.

[11] Shelby Pereia and Thierry Pun, “Fast Robust Template Matching for Affine Resistant Image

Watermarks”, Proceedings of the Third International Workshop on Information Hiding, Lecture

Notesin Computer Science (LNCS), Vol. 1768, pp. 199 - 210, 1999.

[12] N. F. Johnson and S. Katzenbeisser, “A Survey of steganographic techniques in Information Hiding

Techniques for Steganography and Digital watermarking”, pp. 43-78, 2000.

[13] Lin YK. “High capacity reversible data hiding scheme based up on discrete cosine transformation”,

Journal of System Software, 2012.

[14] Chan YK, Chen WT, Yu SS, Ho YA, Tsai CS, Chu YP, “A HDWT-based reversible data hiding

method”, Journal of System Software, vol.82, pp.411–21, 2009.

[15] Yeh HL, Gue ST, Tsai P and Shih WK. “Wavelet bit-plane based data hiding for compressed images”,

International Journal of Electron Communication, 2013.

[16] Meenakshi R, Kuppusmay, “The use of least significant bit technique at various color spaces for

secure steganography with performance evaluation”, International Journal of Advance Research

Computer Science Software Engineering, vol.4, no.6, pp.1047-1051, 2014.

[17] Zebbiche, K. and Ghouti, “Protecting fingerprint data using watermarking”, First NASA/ESA Conf. On

Adaptive Hardware and Systems (AHS‟06), pp.451–456, 2006.

[18] Ateniese, Blundo, De Santis and Stinson, “Visual cryptography for general access structures”, Journal

of Information Computation, vol. 129, pp.86-106, 1996.

[19] Naor, M. and Pinkas, “Visual authentication and identification”, Advances in Cryptology Lecture

Notes in Computer Science, vol. 1294, pp. 322–336, 1997.

[20] Yang, “Visual cryptography: An introduction to visual secret sharing schemes”, Department of

Computer Science and Information Engineering, 2010.

[21] Mohamed M. Ibrahim, Neamat S. Abdel Kaderand M. Zorkany, “Video Multiple Watermarking

Technique Based on Image Interlacing Using DWT”, the Scientific World Journal, 2014

[22] Chitrasen and Tanuja Kashyap, “Digital Video Watermarking using DWT for Data Security”,

International Journal of Advanced Research in Computer and Communication Engineering, Vol. 4,

no.1,2015.

[23] Marwen Hasnaoui and Mihai Mitrea, “Multi-symbol QIM video watermarking”, Signal Processing:

Image Communication, vol. 29, pp.107–127, 2014.

[24] Agilandeeswari Loganathan and Ganesan Kaliyaperumal, "An adaptive HVS based video

watermarking scheme for multiple watermarks using BAM neural networks and fuzzy inference

system", Journal of Expert Systems with Applications, Vol. 63, 30, pp. 412– 434, 2016

[25] N.Mohananthini and G.Yamuna, “Performance optimization of Multiple watermarking using Genetic

Algorithms”, International Journal of Enterprise Network Management, Inderscience Publications

(Article in Press).

[26] Tanima Dutta and Hari Prabhat Gupta, "A Robust watermarking framework for High Efficiency

Video Coding (HEVC) – Encoded video with blind extraction process", Journal of Visual

Communication and Image Representation, vol. 38, pp. 29–44, 2016.

CITE AN ARTICLE

Shankar, T., & Yamuna, G. (2017). AN OPTIMIZED DWT BASED APPROACH IN VIDEO

WATERMARKING WITH MULTIPLE WATERMARKS USING OPPOSITIONAL KRILL

HERD ALGORITHM. INTERNATIONAL JOURNAL OF ENGINEERING SCIENCES &

RESEARCH TECHNOLOGY, 6(9), 611-622.


Documents

ISSN: 2277-9655 et al., IC™ Value: 3.00 CODEN: IJESS7 … /Archive-2017/September... · · 2017-09-25frequency domain watermarking systems are comparatively predominant than the